Entropy guided unsupervised domain adaptation for cross-center hip cartilage segmentation from mri

Abstract

Hip cartilage damage is a major predictor of the clinical outcome of surgical correction for femoroacetabular impingement (FAI) and hip dysplasia. Automatic segmentation for hip cartilage is an essential prior step in assessing cartilage damage status. Deep Convolutional Neural Networks have shown great success in various automated medical image segmentations, but testing on domain-shifted datasets (e.g. images obtained from different centers) can lead to severe performance losses. Creating annotations for each center is particularly expensive. Unsupervised Domain Adaptation (UDA) addresses this challenge by transferring knowledge from a domain with labels (source domain) to a domain without labels (target domain). In this paper, we propose an entropy-guided domain adaptation method to address this challenge. Specifically, we first trained our model with supervised loss on the source domain, which enables low-entropy predictions on source-like images. Two discriminators were then used to minimize the gap between source and target domain with respect to the alignment of feature and entropy distribution: the feature map discriminator DF and the entropy map discriminator DE. DF aligns the feature map of different domains, while DE matches the target segmentation to low-entropy predictions like those from the source domain. The results of comprehensive experiments on cross-center MRI hip cartilage segmentation show the effectiveness of this method.